Transworid Research Network
37/661 (2), Fort P.O., Trivandrum-695 023, Kerala, India
RFSFARCH
‘.
F1W()P K
Thin Solid Films: Process and Applications, 2008: 189-227 ISBN: 978-81-7895-314-4
Editor: S.C. Nam
‘Ti0 thin films for
2
photocatalytic applications
K. Eufinger’, D. Poelman
, H. Poelman
2
, R. De Gryse
2
2 and G.B. Marin
3
Centexbel-Gent, Zwijnaarde, Belgium; 2
Department of Solid State Sciences
Ghent University, Gent, Belgium; 3
Laboratorium voor Petrochemische
Techniek, Department of Chemical Engineering, Ghent University
Gent, Belgium
Abstract
2 is a wide band gap semiconductor which
Ti0
has become well known as photoactivated catalystfor
water and air purfication and self cleaning surfaces.
When illuminated with ultra-band gap ultraviolet (UV)
light, electron-hole pairs are ,generated. To be utilized,
the charge carriers need to migrate to the surface of
the thin film, where they can initiate redox reactions
such as the breakdown oforganicpollutants. The charge
carriers can also become trapped at defect sites in
the material, which enhances their chance of
recombination so that they are lost to the process. As
a result, the photocatalytic activity of Ti0
2 depends
Correspondence/Reprint request: Dr. D. Poelman, Department of Solid State Sciences, Ghent University, Gent
Belgiurn. E-mail: Dirk.Poelinan@UGent.be
190
K. Eufmger et al.
defect
on several factors, inciuding the surface area and the number of
of
states. In turn, these are determined by the stoichiometry, the degree
crystallinity and the crystallite size. Another important factor is the film
a
thickness: Depending on the film morphology and the light source used
of
critical thickness is observed below which there is a strong dependence
the
the photocatalytic activity on the film thickness and above which
dependence is much smaller. The influences of -surface area and film
thickness are ofien underestimated and in many cases even neglected when
e
comparing dijferent thin films. A major issue is also the lack of a suitabl
2 catalysts in the form of
reference thin film, which exists for powder Ti0
Degussa P25.
in
In this paper we will give an overview of the state of the art
2 thin film deposition, with a special emphasis on direct
photocatalytic Ti0
we
current (d.c.) magnetron sputtering as deposition technique. In this context
when
will also discuss important issues that need to be taken into account
comparing dfferent thin films.
1. Introduction
2 as a photocatalytic material was discovered by
The importance of Ti0
of
Fujishima and Honda in 1972, who observed the photocatalytic splitting
2 is activated by
2 electrodes [1]. In the photocatalytic process Ti0
water on Ti0
illumination with (UV) light having an energy higher than the band gap. Given
a band gap of Eg = 3.2 eV for anatase and 3.0 eV for rutile [2] this corresponds
c
to wavelengths below 387 and 413 nm, respectively. The photocatalyti
breakdown reaction proceeds via intermediate steps ending in the
2 and niineral acids. The initial step
mineralization of the organic to water, C0
ns
is the electron-hole pair formation, followed by theirseparation. The electro
es
can be used for reduction, the holes for oxidation processes [2]. The lifetim
d
of the electron and the hole have an influence on how well they can be utilize
2 lattice
for the subsequent redox reaction. Structural imperfections in the Ti0
generate trap sites and recombination centers, leading to a decrease of the
electron and hole concentration [3-5].
In recent years, applications to environmental clean up have been one of
the most active areas in heterogeneous photocatalysis [4-8]. Here one looks at
the photo-degradation of hydrocarbons either in water or in air using
. There are several advantages
2
suspended small particles or thin films of Ti0
2 is
of this material over other photocatalytically active semiconductors: Ti0
as
low cost, chemically inert, absolutely non-toxic, highly photoactive, as well
self-regenerating and can easily be recycled [4]. Additionally, the redox
0/ *OH couple (-2.8 eV) lies within its band gap energy [2],
2
potential of the H
so that it can be used for water splitting.
2 thin films for photocatalytic applications
Ti0
•
-
-
-
-
191
Ti0 has several naturally occurring modifications, the most common
2
ones being rutile, anatase and brookite. Most research with respect to
photocatalysis is performed using anatase [4-8], while in all other aspects
rutile has been much more thoroughly characterized (e.g. electronic
behavior, bulk structural investigations). This is mostly due to the
preparation methods: While it is easier to obtain anatase than rutile at low
temperatures below 600°C, it is impossible to obtain anything but rutile at
temperatures above 800°C, the transition temperature to the
thermodynamically most stable phase rutile [4]. Latter is the case during
standard ceramic processing which requires temperatures of up to 1700°C.
Very littie is known about the brookite phase.
While Ti0
2 powders or films obtained at low temperatures are fmely
grained (nanocrystalline) because little gram growth can occur, gram growth at
temperatures above 8 00°C can be substantial. Therefore, depending on the
preparation method, the nature of the Ti0
2 particle or film surface, as well as
the active surface area are usually very different for anatase and rutile
powders. Since for a catalyst its active surface area is important, differences in
surface area must be taken into account when comparing the activity of
different materials.
2 powder or thin film requires
Generatmg a larger surface area for a Ti0
reducing the gram size. Most photocatalytic materials are nanocrystalline, i.e.
having a gram or crystallite size below 100 nm. When the crystallite size
decreases below about 10 nm, gram boundaries, which present a defective
lattice, start to dominate the material. Defects in the lattice structure are
detrimental to the photocatalytic activity as they act as trap sites for the
photogenerated charge carriers, increasing their probability of recombination.
This is why the highly defective amorphous state is expected to show no
significant or only a very low photocatalytic activity, [4, 5, 7, 9]. Studies have
shown, though, that Ti0
2 which does not show any X-ray .ffraction pattem
(and is thus called XRD amorphous) shows very small crystalline domains
when investigated with transmission electron microscopy (TEM) [10]. In
general, amorphous Ti0
2 has been studied littie.
Originally, Ti0
2 was applied in powder form for water cleaning. The
difficulty to separate the powder from the water and expanding the
applications to air cleaning resulted in using thin films of immobilized
powder or directly deposited Ti0
. The complete reactor walis can be coated
2
to act as photocatalytically active surfaces, but it has to be ensured that the
complete catalyst surface is illuminated. Additionally, it is not always
practical to coat the reactor wails. Therefore, often a coated sample area is
tested inside the reactor. Since thin films are fixed to the substrate, they will
not have an as large active surface area compared to that of a suspended
powder.
192
K. Eufmger et al.
There are different methods for thm film deposition which can be divided
into two general areas, namely wet chemical and vapor deposition. The wet
chemical technique employed most often is sol-gel [4, 1 1]. In earlier years
the limitations of this wet process were due to the fact that the films had to
be calcined at temperatures between 500-600°C, in order to achieve good
stoichiometry and crystallinity. More recent investigations showed that these
high temperatures were due to the wrong substrate choice (soda-lime glass)
and caused by Na-contamination of the films [12]. Further development of
2
the sol-gel technique has led to. procedures which can yield crystalline Ti0
powders at temperatures lower than 200°C [13]. For good bonding of a sol
gel film to a glass substrate, temperatures above 400°C are still necessary,
though [4].
Chemical vapor deposition (CVD) and physical vapor deposition (PVD),
including electron beam evaporation and (magnetron) sputtering [14] are
frequently used vapor deposition techniques. With CVD powders as well as
thin films can be synthesized, depending on the processing conditions.
Physical vapor deposition techniques are used exclusively for thin film
deposition. The main advantage of vapor deposition techniques, especially
P\7D, is that good adhesion of the film to the substrate can be achieved. A
second advantage is that the process conditions can be controlled to yield
crystalline thin films without extended extemal substrate heating.
2
In this review we will first give an introduction to the activity of Ti0
(seçtion 2) and discuss the important factors determining the photocatalytic
activity of a thin film (section 3). Then we will give an overview of the
2 thin film deposition (section 4) and using
different techniques used for Ti0
.. targets we will
2
the example of d.c. magnetron sputtering from ceramic TiO
2 thin films with a high photocatalytic activity
point out how to deposit Ti0
(section 5). In the end we will formulate some general considerations when
2 thin films (section 6). The issue of
testing the photocatalytic activity of Ti0
doping, which is used to modify the electronic properties and thus the
2 will not be addressed as it is: beyond the scope
photocatalytic activity of Ti0
of this review. A detailed discussion can be found in [15].
2
2. Photocatalytic activity of Ti0
Photocatalytic reactions are a special type of catalyzed chemical reactions
which proceed under the mvolvement of light. When the catalyst is a
, this is called semiconductor photocatalysis. When
2
semiconductor like Ti0
) is activated by light electron-hole pairs are
2
the semiconductor (Ti0
generated, and the electrons and holes can be utilized for electron transfer
(redox) reactions at the surface of the semiconductor. All possible processes
are summarized in Figure 1.
2 thm films for photocatalytic applications
Ti0
193
UV-Iight
Figure 1. Scheme of the photocatalytic breakdown reaction of an organic molecule
(OR) on the surface of a photoactivated semiconductor.
2.1. Photoactivation of Ti0
2
Irradiation of the semiconductor Ti0
2 with light having an energy larger
than the band gap (E Ebg) will generate electron-hole pairs, followed by their
separation:
2
Ti0
hv
>TiO
(
2
e/h)—*eb +hb
(1)
As shown in Figure 1 the separated electrons and holes can migrate to the
sample surface, where the electrons are available for reduction, the holes for
oxidation processes [2]. The electrons are associated with the Ti
4 (forming
), but they are immediately scavenged by adsorbed oxygen, resulting in the
3
Ti
reduction of the latter. As a result, the lifetime of the Ti
3 states is very short,
so that the Ti0
2 is never truly in a reduced state [7]. The holes are associated
mostly with —OH groups present at the surface due to saturation of the
terminating 0 with water molecules from the ambient air. Also O and 03 are
suggested as hole trap sites [4, 6, 7], the latter being formed from chemisorbed
02. The resulting radicals oxidize the adsorbed organic compound(s) while
being reduced themselves. The lifetimes of the electron and the hole determine
their efficiency for the subsequent redox reaction. Structural imperfections in
the Ti0
2 lattice generate trap sites and recombination centers, leading to a
decrease of the electron and hole concentration [3-5].
2.2. Photocatalytic (breakdown) reactions
We now look at the potentials of the electrons and holes generated by
photoactivation in Ti0
. The electrons have a reduction potential reflected by the
2
energy level of the bottom of the conduction band, while the holes have an
oxidizing potential reflected by the top of the valence band. Therefore, the redox
potential of the redox couple to be catalyzed must be situated between these two
K. Eufmger et al.
194
energy levels, i.e. within the band gap of the semiconductor [2,4, 8]. In the case
of reactions in water, the positions of the band edges can be referred to the
normal hydrogen electrode (NNE) [2], which is also a reference for the redox
couples of cijemical reactions in water. The redox potential, though, is influenced
by the pH of the solution, which can also influence the surface of the solid
semiconductor. For reactions in the gas phase no standards such as the NHE
have been defmed. Here, among others, the value of the redox potential
difference will depend on the extent of interaction between the two phases,
inciuding the degree of chemisorption, the gas pressure and humidity.
The photocatalytic breakdown reaction of an organic compound normally
proceeds via intermediate steps, ending in the mineralization of the organic to
, water and mineral acids, summarized by equation (2):
2
C0
2
Organic+0
hv
band gap energy
+mineral acids
+H
0
22
>C0
semicondudor
Mineral acids are generated if there are any hetero-atoms, such as S, N,
and Cl present in the organic. This catalytic process is exergonic (i.e. the free
Gibbs enthalpy AG < 0) and the energy of the photons is not used in the
reaction other than to excite the catalyst. As discussed in the previous
paragraph, the band gap and the position of the band edges of the
semiconductor determine which breakdown reactions it can catalyze [2].
Other examples of photocatalytic ‘breakdown’ reactions are partial
oxidation reactions, of e.g. alcohols to aldehydes or ketones, which can also be
seen as synthesis of new organic molecules. This inciudes photocatalytically
assisted polymerization [16]. A good overview of the different possible
reactions is given in [4]. These processes should not be confused with the
endergonic (non-spontaneous) synthesis (i.e. photosynthesis) of organic
molecules which occurs in biological processes.
3. Structural factors controlling the photocatalytic
activity of thin films
2 film normally consists of crystalline grains which are characterized
A Ti0
by their crystal phase, size, shape and arrangement on the substrate. As a result
it can be expected to show a broad range of defects. In this section we will
2 structure and
discuss the influence of individual defects on the TiO
photocatalytic activity, but as will become dear, many defects do not occur
alone but only together with others. Doping is not a topic of this review article
but we would like to point Out the effect of sodium uptake from the substrate.
The defect state of the thin film, as well as its crystal phase, influences its
electronic nature. As a result the observed optical band gap and the charge
2 thin films for photocatalytic applications
Ti0
195
carrier mobility can be altered, which in turn influences the photocatalytic
activity. The latter also depends strongly on the nature and area of the film
surface, where the catalytic reaction takes place.
3.1. Crystal structure
An important factor for the photocatalytic activity is the crystal structure
of the catalyst. The three most studied polymorphs of Ti0
2 are rutile, anatase
and brookite, with the former two being the most important for photocatalytic
applications. Even though the crystal lattices of rutile and anatase are similar,
there are a few significant differences. While rutile has a direct forbidden band
gap of 3.02 eV, the one of anatase is indirect allowed and has a value of
3.2 eV. Rutile and anatase also have different numbers of active sites [4]. A
common observation is that for anatase always good photocatalytic actïvity is
observed, while rutile shows good photocatalytic activity in some [17] and
almost zero activity in other studies [2, 4]. The small difference in band gap
between rutile and anatase cannot explain their different activity. A main
difference, though, is their recombination rate of photoinduced electrons and
holes, which is higher for rutile [2]. Often heat treatment at high temperatures
is used to obtain the rutile phase, resulting in the irreversible loss of surface
hydroxylation and increased crystal growth, both of which decrease the
photocatalytic activity of Ti0
. On the other hand a combination of the two
2
crystal phases can lead to an enhancement of the photocatalytic activity [1820]. A good example is the P25 powder from Degussa, which has become a
reference powder photocatalyst. According to [18, 20] it is the interface
between the two phases that yields the active sites.
Brookite has been investigated to a much less extent. A study on single
crystal brookite, which had a pale brown color, yielded a band gap of 1.9 eV,
which was claimed to be indirect [21]. Since it is difficult to obtain in pure form,
not enough is known about brookite to compare its photocatalytic activity to that
of anatase and rutile. In [22] brookite powders were synthesized and compared to
Degussa P25. As the band gap is substantially lower than that of anatase and
rutile it can be expected that its photocatalytic behavior will be quite low.
The relative stability of the three most common Ti0
2 modifications is
particle size dependent. Below a crystal size of about 15 nm anatase is the
most stable phase while for crystal sizes above 35 nm rutile is the stable phase.
For the range between 15 and 35 nm rutile and brookite seem to have the same
stability [4, 5].
3.2. Crystal defects
Any real material does not have a perfect crystal structure. Even in a single
crystal one finds point defects (misplaced lattice atoms/ ions, vacancies,
196
K. Eufmger et al.
foreign atoms! ions) and dislocations. In polycrystalline materials defects are
added due to the presence of gram and crystallite boundaries. These defects
can introduce energy levels inside the band gap, their location depending on
the nature of the defect. According to semiconductor theory [23] donor states
are found just below the conduction band, while acceptor states are found just
above the valence band. Defect levels deep inside the band gap are trap levels,
both for electrons as well as for holes. In photocatalysis literature often the
term ‘trap site’ is used, which denotes a site that can permanently (deep) or
temporarily (shallow) bind a charge carrier. The definition of a trap site is not
straightforward, though, since all levels inside the band gap have the potential
to be trap sites. The deeper the trap site lies within the band gap the higher its
trapping power, i.e. the lower the chance that the trapped charge carrier can
leave the trap withm its lifetime. But also the donor and acceptor levels close
to the band edges can act as trap sites. The latter enhance the separation of the
electron-hole pairs, but if the charge carriers are too strongly bonded they are
not available for the photocatalytic redox reaction. As a result, the activity
enhancement or reduction by a defect depends strongly on its nature, and the
presence of other defects.
3.2.1. Crystallinity
The degree of crystallinity was found to be an important factor influencing
the photocatalytic activity of Ti0
. It depends on the actual crystallite size and
2
the ratio between crystalline and amorphous phase present in the material. A
higher degree of crystallinity resuits in a lower number of crystal defects,
which is beneficial for the photocatalytic activity of Ti0
. Note that gram size
2
and crystallite size may not be identical. The crystallite size is defined by the
domain (within a gram) having the same crystallographic orientation, while a
gram is defined by the presence of (visible) gram boundaries. Therefore, it is
possible to have different crystallites within one gram.
When trying to asses the effect of crystallite size on the photocatalytic
activity one faces as main difficulty how to change the crystallite size while
keeping the gram size constant. One needs to increase the temperature to
promote crystal growth in anatase Ti0
, which normally resuits in gram
2
growth and possibly transformation to the rutile structure. Careful studies
using Si0
2 as gram growth inhibiting additive have shown that a higher degree
of crystallinity resuits in a higher photocatalytic reactivity [24-26]. The gram
growth inhibiting properties of Si0
2 also retard the anatase to rutile
transformation which allows annealing at higher temperatures, resulting in
improved crystallinity [4]. When decreasing the gram size below ca.
10 to 20 nm the degree of crystallinity is severely reduced due to the high
surface to bulk ratio, since the surfaces (gram boundaries) represent a high
defect state.
2 thin films for photocatalytic applications
Ti0
197
The “amorphous Ti0
” phase represents the highest state of disorder, i.e.
2
has a completely random structure [11]. With Ti0
2 the definition of amorphous
is difficult, as there is always some degree of local ordering observed [10].
Here we will define “amorphous Ti0
” as the phase where no signal can be
2
detected by x-ray diffraction (XRD), which means that the material has
crystalline domains of a size below the detection limit (<3-5 nm). Usually
grains around 5 nm are detected as very broad peaks around the position
expected from the diffraction spectrum of the material and such materials are
often labeled ‘nanostructured’ [27]. This can be confusing as the usual
definition of the nanoscale is < 100 nm [28].
Very littie is known on the electronic structure of amorphous Ti0
. Most
2
discussions agree that the high defect concentration should result in poor
charge carrier conduction [5, 13]. In [5] it is claimed that for such materials
(which they count as nanocrystalline Ti0
) one observes a direct forbidden
2
band gap transition as compared to the indirect allowed transition observed in
well crystallized anatase. Completely amorphous Ti0
2 is expected not to show
any photocatalytic activity [13, 16]. As a result, only rarely the photocatalytic
activity of amorphous films or powders is measured [13, 29]. In our own
studies we have found that XRD amorphous Ti0
2 films can have a substantial
photocatalytic activity, though, depending on their microstructure [30]. The
photocatalytic activity of such films increases by a factor of almost 3 when
crystallizing anatase [30, 31], confirming that a higher degree of crystallinity is
indeed beneficial.
3.2.2. Oxygen vacancies (stoichiometry)
Oxygen vacancies represent a special type of defect in Ti0
. They are
2
present when the material is not completely stoichiometric or they can be
generated by impurities. As a result, oxygen vacancies will not occur on their
own but only in combination with one or more other defects (e.g. Ti
,
3
impurity). The effect of oxygen vacancies will vary, depending on the exact
circumstances of their occurrence. This is reflected in the seemingly
contradictory experimental resuits obtained by different researchers.
Indications that oxygen vacancies are detrimental to the photocatalytic
activity were found by [11] and [32] for d.c. magnetron sputtered Ti0
2 films.
180
Post oxidation of films using
tracer and subsequent SIMS (secondary ion
mass spectroscopy) analysis showed that with increasing amount of
incorporated tracer the photocatalytic activity decreased [11]. From this, the
decrease in photocatalytic activity was attributed to the increase in the number
of defects associated with oxygen vacancies in the film. In [32] the
photocatalytic activity was linked to the thin film stoichiometry determined by
ESCA (electron spectroscopy for chemical analysis) showing a decrease in
photocatalytic activity with decreasing oxygen content. In both cases the
198
K. Eufmger et al.
conciusion was that for optimum photocatalytic perfo
rmance the films must be
fully stoichiometric.
In [33] preparation of a visible light active photocata
lyst of reduced Ti0
2
is claimed. The color of the catalyst is reported to be
yellow and traces of N
were found, indicating doping. In [34-36] it is reported
that the doping of Ti0
2
thin films or powders with N resuits in visible (VIS
) light photocatalytic
activity, which could explain the results obtained in
[33]. In a different study
by [35] commercially available anatase powder was redu
ced in a H
-plasma to
2
have an 0! Ti ratio of 1.68 (ESCA estimate). This
powder had a yellowish
color and showed a red shift of the optical absorptio
n edge and VIS light
photocatalytic activity. An electron spin resonance
(ESR) signal of g = 2.003
found by the authors was assigned to electrons trapped
on an oxygen deficient
site [35]. On the other hand, [37] showed that H was
incorporated into Ti0
2
thin films sputtered in a hydrogen atmosphere. The result
was the formation of
deep levels inside the band gap. These authors did
not investigate the
photocatalytic activity of their films, but the resuits
indicate that also in the
study presented in [35] the doping with H interferes with
the role of the oxygen
vacancies.
Density of states (DOS) calculations performed by
[11] showed no effect
of single oxygen vacancies, but for double oxygen
vacancies energy levels
inside the band gap were generated. It was suggested
that these energy levels
can act as rëcombination centers for the charge
carriers, decreasing their
concentration and, therefore, the photocatalytic activity
of the material. On the
other hand, band gap calculations by [38] for anata
se showed that oxygen
vacancies introduced shallow donor states in he band
gap. The authors expect
these to increase the photocatalytic activity of Ti0
.
2
In conclusion, the effect of oxygen vacancies on the phot
ocatalytic activity
of Ti0
2 is quite complex. The experimental resuits indic
ate that the presence of
oxygen vacancies alone is not beneficial [11, 32],
while a positive effect is
found when they occur together with secondary
defects due to N or H
incorporation [33, 35].
3.2.3. Lattice strain
1f the size of a defect is different from its host, a
lattice distortion (i.e.
strain) is introduced. Some early studies looked at the
influence of an applied
external stress on the band gap transition of Ti0
2 (rutile) [39]. It was found
that the band gap increased with increasing applied (unia
xial) stress along the
a-axis, yielding a pressure coefficient of dEg/ dP of
-1 .19 1 06 eV/ bar (note
that the negative sign indicates compressive stress).
In [40] the high band gap of 3.4 eV observed
for magnetron sputtered
anatase Ti0
2 thin films as compared to the bulk value of 3.2
eV was attributed
to the distortion of the Ti0
2 unit ceil. Related authors also performed
4
2 thin films for photocatalytic applications
Ti0
199
experiments and ab-initio calculations [41, 42] to determine the influence of
lattice distortions (i.e. lattice strain) on the band gap of rutile and anatase Ti0
.
2
The increase of the band gap with decreasing lattice constant a and increasing
lattice constant c was confirmed by experimental resuits of these authors. From
these resuits one can extract a value of -4.89 106 eV/ bar for the dependence of
the band gap Eg on an applied uniaxial stress along the a-axis [15], which is on
the order of the one for rutile (see above). This shows a good agreement
between the stress measurements and the strain calculations, considering that
rutile has a higher Young’s modulus than anatase [43].
3.3. Sodium contamination
Early studies discovered that the photocatalytic activities of annealed Ti0
2
thin films depended strongly on the type of substrate used [44]. It was
suspected that ions diffusing from the substrate into the film during annealing
influenced its activity. Other authors investigated the effect of a soda-lime
glass substrate on the photocatalytic activity of heat treated Ti0
2 films
[12,46].
In [12] the resulting decrease in photocatalytic activity was related to the
effect of Na on the gram growth in the thin film upon heat treatment. Na was
found to retard the crystallization of anatase while increasing the particle size
in the film. The photocatalytic activity was correlated to the particle size of the
films, rather than to their Na content, the photocatalytic activity increasing
with decreasing particle size (in the range of 25 to 40 nm). Na was expected
to form a shallow donor state in Ti0
, which would make it a rather unlikely
2
recombination center. Therefore, the authors conciude that Na in the films
influences the photocatalytic activity by changing the particle size and not by
acting as recombination center.
Experiments in our group have shown that Na inhibits crystallization by
decreasing the degree of crystallinity and increasing the temperature needed
for crystallization [46]. The photocatalytic activity was not only decreased for
crystalline films, but also for films annealed below the crystallization
temperature, which remained amorphous and in all cases the gram size
remained constant [31, 46]. Latter indicates, in contrast to [12], that it is indeed
the uptake of Na into the films that degrades the photocatalytic activity. In
[45] the diffusion of Na into the Ti0
2 thin film from soda-lime glass was
measured by ESCA depth profiling and a direct link between its lower
photocatalytic activity and the Na-concentration was made. It was suggested
that Na acted as recombination center for electrons and holes.
Independent of the exact cause, the uptake of Na decreases the
photocatalytic activity of Ti0
2 thin films substantially. As a result most
investigators use soda-lime free substrates or a Si0
2 barrier coating [11, 47] to
200
K. Eufinger et al.
avoid incorporation of Na when a heat treatment step is needed after
deposition. Uptake of Si from any kind of glass is observed, but this is not
2 thin
reported to have a negative influence on the photocatalytic activity of Ti0
films [45]. In conciusion, the best strategy concerning the effect of Na on the
2 is to avoid the possibility of Na-contamination by the
properties of Ti0
correct choice of (glass) substrate.
3.4. Surface hydroxylation
The surface hydroxyl groups are important shallow trap sites for the holes
2 (section 2.1). Therefore, they play a
generated upon UV-illumination in Ti0
crucial role in the photocatalytic process. The degree of surface hydroxylation
not only depends on the relative humidity but also on the surface structure of the
catalyst. Hydroxyl groups are formed by water dissociation, so that it is favorable
to have paired acid! base sites situated at the appropriate distance [4]. The water
molecules are initially bonded to the surface sites with an acid character, namely
Ti-cations with a coordination number lower than the bulk value of 6. Adjacent
surface sites with a basic character, namely bridging oxygen atoms, then accept
the proton. The conciusion is that the degree of surface hydroxylation increases
with the number of acid sites present at the surface. The coordination number of
Ti at the surface depends on the terminating crystal facet [6] but also decreases
with its oxidation state. Therefore, sub-stoichiometric surfaces show an increased
surface hydroxylation [4]. Si also has a lower coordination number of 4 (bulk)
2 increases the degree of surface
and it is reported that its addition to Ti0
hydroxylation [48]. Consequently, the catalyst preparation conditions play a
major role in surface hydroxylation [4,49]. The degree of surface hydroxylation
in a gas atmosphere is related to its water content, while in solution it is related to
its pH. Therefore both factors have an influence on the photocatalytic reaction
rate [4].
3.5. Particle size
In general, particles in the nanometer range (<100 nm) show a physical
and chemical behavior different from larger sized material. For very small
particles (<35 nm) their size has an effect on the crystallographic phase
stability and the chemical activity (section 3.1). It has become common to cail
such materials nanostructured, with nanoparticles being ultrafinely dispersed
particles and nanocrystalline materials being materials with a fine crystallite or
gram size. Unfortunately the terms ‘crystal’ and ‘gram’ are sometimes used as
synonyms even though they describe something completely different
(section 3.2.1).
2 there is quite
Regarding the photocatalytic activity of nanocrystalline Ti0
some discussion on the influence of the crystallite and particle size. The degree
2 thm films for photocatalytic applications
Ti0
201
of crystallization (i.e. the ratio between crystalline and amorphous phase) will
decrease with decreasing gram size due to the increasing mfluence of the
(amorphous) gram boundaries. Often the gram size and! or crystallinity of the
thin films are not clearly defined or have a large spread, so that dependencies
cannot be defined well.
In the following we highlight the most important effects that are associated
with a nanoscale structure in Ti0
. So far, no solid proof could be provided
2
that the general photocatalytic efficiency of very finely grained (< 10 nm) Ti0
2
(signifïcantly) exceeds that of larger grained material, as long as the gram size
remains below 100 ilm. The photocatalytic activity of Ti0
2 having a (lateral)
gram diameter above 1 I.Lm has been studied littie.
3.5.1. Chemical activity
With nanostructured materials the particle surface area is large compared
to the bulk of the particle. As a result the surface energy becomes more
dominant, changing the Gibbs free energy of the material and, therefore, its
chemical activity. On the other hand, the large surface area itself increases the
catalytic activity of the nanostructured material by offering more active sites
for the reaction to take place. The surface presents a defect state due to the
presence of dangling bonds on the Ti0
2 surface, which is sometimes
interpreted as an increased number of oxygen vacancies. It is argued that this is
why nanosized Ti0
2 will also have a higher density of oxygen vacancies than
large gram size Ti0
2 [5]. These defect states are preferred sites for the
adsorption of water and 02 from the ambient air, the former resulting in the
formation of hydroxyl groups (section 3.4).
3.5.2. Optical and electronic properties
The fact that the gram dimension is smaller than the wavelength of visible
light (ca. 400-700 nm) also resuits in different interaction with such light.
Compared to films with a larger gram size, which can be opaque, one observes
that the degree of light scattering is decreased [8, 50]. This means that the
films remain optically transparent. The small gram size also has the advantage
that the charge carriers do not have to migrate far before they reach the surface
of the gram, where they are needed for the redox reaction at the catalyst
surface (Figure 1).
On the other hand, the small particle size is associated with a higher
number of defects in the form of surfaces [4], while the number of inner gram
defects is decreased [51]. Defects act as trap sites, so that the number of trap
sites decreases in the bulk but due to the increased surface area a larger number
of surface trap sites is generated. When the ratio of surface to bulk reaches a
critical value the increasing number of trap sites can offset the positive
influence of the short carrier mgration distance. As a result, there is a critical
202
K. Eufmgeretal.
size below which the surface recombination of electrons and holes becomes
dominant due to the increased surface to volume ratio. The optimum gram size
is reported to be around 10-20 nm [52], below which a decrease in
photocatalytic activity is observed.
3.5.3. Quantum size effect (very small gram sizes)
For decreasing particle sizes, quantum mechanical calculations show a
transition from semiconductor to molecular properties. This means that the
band structure of the semiconductor breaks up into quantum levels, the
transition point being where the particle size reaches the order of the de
Broglie wavelength of the charge carriers in the semiconductor. For most
semiconductors this lies in the range of 5-25 nm, the exact value depending on
the material [8]. This change in electronic structure is commonly referred to as
‘Quantum size effect’ (QSE) or ‘Q-effect’ and the respective particles are
called ‘Q-particles’. The most wide-spread, simple theoretical framework to
study the influence of confinement effects and how they depend on the primary
particle size is the so-called effective mass approximation (EMA). More
comprehensive discussions of the quantum size effect and the EMA are given
in [2, 5, 7, 8, 53, 54]; the original theoretical framework was developed by
Brus in his work on CdS clusters [55].
Strictly speaking the theory behind the quantum size effect is only valid
for (i) freely suspended particles, where the charge carriers are confined within
the particle and (ii) a covalent material, where the electronic band theory is
2 thin film,
valid [55]. Both conditions are not fulfihled in the case of a Ti0
where the bonding is partially ionic and the particles are embedded, allowing
charge carrier migration across the gram boundaries. Latter is also valid for
strongly aggiomerated powders where charge carrier transport is no longer
limited to the individual grains but to the (much) larger aggiomerate. The
2 [5, 7, 55], even
quantum size effect could not be verified for (anatase) Ti0
though claims have been made that it is responsible for the blue shift of the
band gap in nanocrystalline anatase powder [54].
3.5.4. Nature of the band gap transition
There is some discussion in literature about the nature of the band gap
. Some authors
2
transition in nanostructured (anatase) and amorphous Ti0
[5, 54] claim that for small gram sizes the band gap transition switches from
indirect to direct, which is of interest as the absorption at threshold and the
emission of light is much stronger for the direct transition. This should result in
a more efficient absorption of light and in improving the photochemical
performances of the nanostructured materials.
In [54] it was decided on basis of the best fitting resuits to the
experimental data that the band gap of very finely grained anatase was direct.
2 thin films for photocatalytic applications
Ti0
203
The authors suggest that photoluminescence studies could be used to support
the observations. In [5] it is argued that the confinement of charge carriers to a
limited space causes their wave functions to spread out in momentum space, in
turn increasing the likelihood of radiative transitions for bulk indirect
semiconductors. Again, this argument is not valid for thin films and
agglomerated powders where the charge carriers cannot be strongly confined
(3.5.3). In [52] the enhancement of absorption observed in small nanocrystals
is claimed to be the result of the very high surface to volume ratio. Here the
share of surface atoms is sufficiently high to increase the chance of surface
absorption, which increases the light absorption in total. Both arguments are
actually based on the fact that a direct transition shows a stronger light
absorption than an indirect transition.
3.6. Microstructure (film surface area)
There are few consistent studies which investigate the effect of the thin
film microstructure on the photocatalytic activity. Mostly the film crystallinity
is studied but the gram structure and porosity are generally ignored. This is
surprising because especially the film porosity will influence the active surface
area, which in tum has a strong effect on the photocatalytic activity. Our own
studies showed that the film microstructure is strongly related to the
photocatalytic activity of the Ti0
2 thin films [30, 31].
It is usually impossible to determine the BET (Brunauer-Emmet-Teller
absorption isotherm) surface area of a thin film because the total surface area
of the sample is too small for the standard BET analyzers. These have a
detection limit for the total surface area in the order of 0.1 -1 m and receptacle
sizes in the cm
3 range. Additionally the receptacles are designed to analyze
powder samples, so that coated substrates have to be cut into pieces, which
resuits in a high amount of uncoated surface being tested. Note that the surface
area of a (powder) catalyst is usually reported in m
/g, which is not applicable
2
for a thin film catalyst where the weight of the thin film is normally more than
3 orders of magnitude lower than that of the substrate. One method to estimate
the active surface area of a thin film is by looking at its microstructure. Pores
will add to the surface area of a thin film (section 3.6.2), but their size will
determine whether they can add to the active surface area (section 3.6.3).
3.6.1. Deternunation of the thin film porosity
A simple way to estimate the porosity is through the refractive index of the
thin film, which can easily be detennined from ellipsometric or optical
transmission measurements [56, 57]. The refractive index of a mixed
compound can be calculated from the indices of the individual components,
using e.g. the Lorenz-Lorentz equation [56]:
204
K. Eufmger et aL
q
mi’ 2
n 2 , ,,
s—, 1 fl•
i
2
—=2
n +2 p ,=
n +2
1
.
—
—
!_Lj
p }
m
with t=1
1=1
where n is the refractive index of a composite material of average molecular
weight M and density p which can be viewed as consisting of m materials each
having the mole fractionj, the molecular weight M, and the density p.
A porous material can be modeled as a mixture with air. Since the
refractive index of air is close to 1, equation (3) simplifies to:
n?+2
n —1 n +2
(4)
with P denoting the porosity, defined as
(5)
Expression (5) allows calculation of the film porosity from the refractive index
if there are no other influences like second phases or dopants. Nevertheless,
any calculation of the porosity should be taken as an estimate only.
3.6.2. Estimate of the surface area of a thin film from its porosity
When depositing a thin film by d.c. magnetron sputtering under the
conditions that yield porous zone 1 growth (please refer to section 5.2)
one
obtains irregular and separated column-grains (Figure 2).
These column-grains have rounded tops and an elongated, approximatel
y
cylindrical shape. The rounded tops will in all cases increase the surface
available for catalysis. The side surfaces of the column-grains can only partici
pate
Figure 2. Scanning electron microscope (SEM) images of Ti0
2 thin films deposited in
zone 1 structure by d.c. magnetron sputtering (60 W, 1.4 Pa,
2 ceramic target):
TiO
top view, b) cross section.
a)
2 thin films for photocatalytic applications
Ti0
205
when the inter-gram spacing is large enough to allow diffusion of the reactant
and reaction products. This inter-gram spacing actually defines the pore size in
the thin film. In the following we shali first calculate the increase in the thin
film surface area due the microstructure and then estimate the pore size for a
given gram size, film thickness and density.
1f a thin film has a completely dense structure and a flat surface, the active
surface area is equal to the geometrical area of the substrate (Figure 3., left). A
rough surface, on the other hand, will have a larger area. We can estimate this
increase by assuming that the surface consists of a collection of semi-spheres
(the tops of the individual grains, see Figure 3., right). A circle with radius R
will have a flat surface area of ‘rR
, but a semi-spherical surface area of 2 irR
2
.
2
As a result, the gain in surface area when going from a flat to a rough surface,
consisting of half spheres, is only a factor of 2.
Much more effective area can be gamed if the film consists of free
standing columns (Figure 3., right). For the sake of simplicity we assume that
every column has a flat top and a fixed square column section, with size d
(Figure 4). The height of the columns is identical to the film thickness h.
Porosity is taken into account by separating the columns by a distance f
(Figure 4).
Thinfilm
Thin film
Iflflflflfl[]flflE
Substrate
S ubstrate
Figure 3. Sketch of a thin film (in cross section) consisting of (spaced) column-grains
(right), illustrating that the dense structure normally assigned to a thin film (left) is not
necessarily correct.
:4—x
4—
d
:
—+
1
f
1
4-
Figure 4. Sketch of the column-gram thin film structure (top view) showing the cdl
used for calculation of the surface area and the free space between the column-grains.
K. Eufmger et al.
206
The projected (geometrical) surface area, equal to the substrate area
(which is assumed flat) is (d +J)2 per gram. On the other hand, the total column
surface area, inciuding the column sides, is:
(6)
4dh
d
+
2
The ratio of the active surface area of the free-standing columns to the area of
a dense film will be:
d
4dh
+4dh
2
d
+
2
d
2
(d+f)
(7)
sincef« d (see Table 1.) for realistic thin films.
1f we take a film thickness of h = 400 nm and a column diameter of
d = 50 nm, this ratio is 33, so that the total active area increases considerably
when the film consists of free-standing columns.
3.6.3. Estimate of the pore size of a thin film from its porosity
The porosity P can be defmed as the fraction of free space in the film, and
is a value which can be estimated from the effective refractive index of the
films according to equation (4). 1f the columns themselves are completely
dense, then we can relate the porosity P to the spacing fbetween the column
grains (Figure 4.):
(8)
2
(d+f)
Solving forfyields:
1
—1
f=d(
J(1-P)
The calculated spacing fbetween two column-grains for a given porosity P is
listed in Table 1. Note that these values only give an estimate since they were
calculated taking the column-grains as 100% dense. Since this is most likely
not the case the true inter-column spacing will be smaller.
The inter-gram spacing f is nothing other than the pore size of the thin
film. According to the IUPAC (International Union of Pure and Applied
Chemistry) definition one distinguishes the following different pore types
according to their diameterx [58]:
macropores: x> 50 nm
mesopores: 2 x 50 nm
micropores: x < 2 nm
2 thm films for photocatalytic applications
Ti0
207
Table 1. Calculation of the inter-gram
spacing .f from the porosity according to
Figure 4.
From Table 1. one can see that from a porosity of about 7.5% on the thm
films are mesoporous. Usually micropores are too small to be catalytically
active, while mesopores are known to increase the active surface area of a
catalyst. The minimum size a pore needs to have in order to be active depends
on several factors. These inciude the diffusion coefficient of the reactants into
and products out of the pores (determined by their size and electronic nature),
their concentration in the atmosphere surrounding the catalyst and the flow rate
of the carrier gas stream passing over the catalyst bed.
Our own resuits have shown a strong dependence of the photocatalytic
activity of column-grained Ti0
2 thin films on porosity values up to 22.5%,
when measuring the breakdown of ethanol in a batch reactor [59]. Increasmg
the porosity above this value no longer had an influence on the photocatalytic
activity, indicating that a critical pore size was reached which allowed the
breakdown reaction of ethanol to proceed freely.
3.7. Film thickness
Not much is reported on the effect of thickness on the photocatalytic
activity of thin films [12, 45, 60], even though some authors are aware of the
influence and correct for the variations in sample thickness [13]. In order to
study the true influence of film thickness it is important that the structure of the
film does not change with film thickness, which was not the case in [60]. In
[45] thin films with thickness between 50 and 250 nm were studied. Here it
was found that the photocatalytic activity reached its maximum at ca. 140 nm,
after which it remained constant. In [12] a critical film thickness of ca. 360430 nm is reported. The difference between the two studies mdicates that this
critical film thickness is dependent on the thin film preparation techniques
and/ or the experimental setup for testing their photocatalytic activity. As we
have shown in section 3.6.3 the density of a column-grained structure
determines its inter-gram porosity, and as a result its surface area. The catalytic
activity of these pores will depend on the experimental setup, including the
size and nature of the organic test substance used. Neither [12] nor [45] report
anything on the microstructure of the thin films studied.
Tada et al. tried to determine the nature of this critical film thickness and
proposed that there is a limited diffusion length of the charge carriers and a
K. Eufinger et al.
208
cumulative light absorption with increasing film thickness [45], as depicted in
Figure 5. Their experimental resuits indicated that the critical film thickness
was ca. 100-150 nm, from which they deduced a charge carrier diffusion
length of about 300 nm.
2 thin films (22.5% porosity) with
Our own experiments using porous Ti0
a column-gram structure have confirmed this reasoning. The data showing the
dependence of the photocatalytic activity on the film thickness (Figure 6.) can
be fitted at first sight by two linear graphs (dark). On the other hand,
an exponential fit describing the light absorption in the film (at an intermediate
wavelength of the given lamp spectrum) can also be applied (light, dotted
curve).
a)
b)
UV-light
J]jJfj Thin film
Thin film
.
Charge carriers
Ecn
oc
Substrate
S u bstrate
Figure 5. Sketch showing the limiting factors for the photocatalytic activity of thin
films: (a) increasing light absorption, (b) the charge carrier diffusion length.
1
““•..•;
1
.
14
..
1
ê
0.8
-1___
1.2-,
/
1
1
,/
gO.6
o
O8
.
•/
:/
0.4
0.6
absorp. fit
/
0.2
,
.1.6
.
/
—
—
linear fit
4’ reaction rate
04
,..
0.2
n
0
200
400
600
800
0
1000
film thickness (nm)
Figure 6. Dependence of the photocatalytic activity on the film thickness for XRD
2 ceramic
2 thin films deposited by d.c. magnetron sputtering (TiO
amorphous Ti0
which seem
sections
linear
two
by
target, pure Ar at 1.4 Pa, 60 W). The data are fitted
to indicate a possible critical film thickness of about 350 nm, and an exponential fit
describing the absorption of UV-light with increasing film thickness.
2 thin films for photocatalytic applications
Ti0
209
Thin film
Substrate
Figure 7. Sketch showing the possible diffusion paths of photogenerated charge
carriers in column-grained Ti0
2 thin films.
In order to determine which of the two mechanisms is correct we
performed separate photocatalytic measurements for two samples of different
thickness using (i) the complete spectral range of the lamp and (ii) only the
wavelengths above 345 nm. The resuits indicated that the higher wavelength
light, which enters deeper into the thin film, showed a similar photocatalytic
activity per active photon arriving at the film surface. This should not be
observed if the diffusion length of the charge carriers was the limiting factor.
Additionally the column-gramed film structure (Figure 3.) also explains,
why the carrier diffusion cannot be the limiting factor for these porous films.
In Figur 7. possible diffusion paths of the charge carriers in the column-grains
are shown. Since the diameter of the columns is only about 50 nm, charge
carriers generated anywhere inside the gram do not need to diffuse more than
about 25 nm to reach its surface. As a result, the limiting factor that the
thickness imposes on the photocatalytic activity is the fact that most of the
incoming light is absorbed in the first few 100 nm of the film. Increasing the
film thickness does not lead to much more absorption. This allows to use the
linear expressions from Figure 6. to determine a ‘critical value’, above which
the photocatalytic activity depends less strongly on the film thickness.
The conciusion is that the film thickness is an important parameter for the
photocatalytic activity of Ti0
2 thin films. Care must be taken when designing
experiments to find the critical film thickness above which the photocatalytic
activity is less dependent on this factor to avoid experimental errors. As
already discussed, the active pore size depends on the chosen organic test
substance, which can be expected to influence this critical film thickness.
3.8. Designrng a highly photoactive Ti0
2 thin film: Summary
To summarize sections 3.1 to 3.7, we can state that to optimize the
photocatalytic activity of a Ti0
2 thin film one has to control two parameters,
namely the surface area and the defect structure, which is primarily reflected in
its crystallinity.
K. Eufinger et al.
210
be increased by decreasing the gram
• The surface area of the thin film can
nt for finding the minimum pore size
size and increasing the porosity. Importa
of the organic substance to be broken
will be the dimensions and the nature
down.
eases by increasing the number of
• The crystallinity of the thin film incr
the gram surface can be taken as a
crystallites as well as their size. Since
crystallinity increases witb gram
highly defective, non-crystalline phase the
size (lower surface to volume ratio).
size (high surface area) and
The result is that one needs to balance gram
optimum resuits. In general it is
crystallinity (low defect state) for the
low as possible.
interesting to keep the number of defects as
2 thin film
ination of the Ti0
Additionally, it is necessary to avoid contam
the substrate.
by inward diffusion of Na or other ions from
es
4. Ceramic thin film deposition techniqu
4.1. General considerations
2
ues used for Ti0
can roughly
niq
The variety of thin film deposition tech
chemical and vapor deposition. The
be divided into two areas, namely wet
the substrate is either coated with a
former is a solution based technique where
2 thin film is grown onto the substrate
colloidal solution (sol-gel) or a Ti0
deposition various techniques are
(electrochemical deposition). In vapor
Ti0 In the
.
some precursor or 2
employed to generate a vapor phase of Ti,
te.
2 film is formed by oxidation at the substra
former two cases the Ti0
substrate material. It must be
An important issue is also the choice of
to remain inert and undamaged
compatible with the deposition conditions as
be resistant to the temperatures
also
st
mu
te
stra
sub
The
.
ion
osit
dep
ing
dur
the thin film and not cause any
required for calcination! crystallization of
nt example are polymer substrates
contamination (see section 3.3). An importa
ce (typically 150°C), and can pose
which have a rather low temperature resistan
ssing of e.g. water.
a problem in vacuum chambers due to out-ga
4.2. Wet chemical techniques
ues employed for thm film
The most important wet chemical techniq
osition, with the former being by
deposition are sol-gel and electrochemical dep
briefly discuss the mam aspects of
far the most frequently used [4, 11]. We will
nded literature for further reading.
these two techniques and refer to the more exte
4.2.1. Sol-gel
nd advantages for thin film
The sol-gel technique has some profou
ibility and ease of processing.
deposition, namely purity, homogeneity, flex
2 thm films for photocatalytic applications
Ti0
211
Depending on the type of precursor used, one distinguishes between the non
alkoxide and the alkoxide route. Former uses inorganic Ti-saits, while latter
uses metal-organic (alkoxides) as precursors. The first step in the sol-gel
process is the formation of a colloidal suspension of solvent stabilized titanium
oxide particles (sol) by controlled precipitation. The sol is then coated onto the
substrate by different techniques like dip or spin coating. During drying
(evaporation of the solvent) a ge! is formed, which has a very loose structure.
Any remaining inorganic salt must be carefully removed by washing of the gel
film.
The route employed most is the one using alkoxides (generic formula O-R,
with R denoting an organic rest). The most popular ones are the tetra-ethoxide,
iso-propoxide and n-butoxide of Ti, with the general formula 4
Ti(O-R) Here
.
the sol is formed by controlled hydrolysis followed by poly-condensation of
the metal-organic precursor. In order to have good control of the process one
aims to use precursors that allow for separating and slowing down the two steps.
For both routes the structure of the thin film can be influenced by the
nature of the precursor and the solvent as well as by adding surfactants. Since
the film thickness achieved by one coating cycle is below 100 nm, several
coatingcycles are needed to obtain thicker films. A very good overview on
depositing sol-gel films of Ti0
2 can be found in [4].
In earlier years the limitation of the sôl-gel process was the necessity to
calcine the films at temperatures between 500-600°C, in order to achieve good
stoichiometry and crystallinity [12] which limited the applicable substrate
materials. More recent investigations showed that the need for such high
temperatures was due to the wrong substrate choice and caused by Na
contamination of the films (section 3.3). For good bonding of a sol-gel film
with a glass substrate temperatures above 400°C are stil! necessary, though [4].
4.2.2. Electrochemical deposition
Electrochemica! deposition, also referred to as electrolysis or
electroplating, is a technique which has been known for a long time. Here the
2 thin film is deposited onto a metal electrode from a solution of a Ti
Ti0
compound like TiC1
4 [61, 62], TiC1
3 [62, 63], Ti(S0
2 [64-66] or
)
4
(NH
[
]
)
4
0
2
TiO(C
4) [67]. Care must be taken that the chosen salt does not
hydrolyze in the solvent used, which would result in precipitation (i.e. powder
formation). Often hydrogen peroxide is added as oxidant in combination with
the Ti-compound [61, 64, 66]. Surfactants can be added to influence the
growth and structure of the deposited film [62, 68]. 1f the substrate is
conductive a d.c. process can be used [62, 64]. As the deposited coating is
insulating, often a pulsed d.c. [32] or a.c. [67] process is necessary. The latter
two also have the advantage that they can be used for coating an insulating
212
K.Eufmgeretal.
substrate or to fl11 the porous structure of an alumina film grown by anodic
oxidation [32, 67].
In most cases a porous, XRD amorphous Ti0
2 film is deposited, which is
beneficial for catalytic applications. Annealing at temperatures between 300
and 500°C is needed to obtain crystalline anatase films [61, 66, 69]. Some
references report direct deposition of crystalline Ti0
2 [62].
A related technique is the anodic oxidation of Ti-metal, where a thin oxide
film is grown on the surface of the anode. There are few reports about growing
such films for photocatalytic applications [70].
4.3. Vapor deposition techniques
Vapor deposition techniques inciude chemical vapor deposition (with or
without plasma enhancement, i.e. CVD or PEC\TD) and physical vapor
deposition (PVD), inciuding electron beam evaporation and (magnetron)
sputtering. With (PE)CVD powders as well as thin films can be deposited,
depending on the processing conditions. Physical vapor deposition techniques
are used exclusively for thin film deposition. The main advantage of vapor
deposition techniques, especially P\JD is that good adhesion of the film to the
substrate can be achieved. A second advantage of vapor deposition techniques
is that the process conditions can be controlled to yield crystalline thin films
without extemal substrate heating, while substrate self heating can occur
during the process.
4.3.1. Chemical vapor deposition techniques
Chemical vapor deposition (CVD)
Chemical vapor deposition (CVD) techniques are based on the oxiclative
decomposition of a volatile Ti-precursor to form a Ti0
2 film on the substrate
(Figure 8.). Variations of the technique inciude using an aerosol, which is
formed in the reaction chamber, so that droplets are deposited. 1f a thin film is
desired, care has to be given, that the process does not result in powder formation.
Ti-precursor
reaction zone
0-precursor]\
0
0’
*Q
carrier gas,
precursors
-.
0
•
0
•0
0
•
0
—
exhaust)
°
•
.
substrate
0
0
0
0
0
Figure 8. Scheme of a generic chemical vapor deposition process.
2 thin films for photocatalytic applications
Ti0
213
A study showing the factors determining powder or film formation is given in
[71]. In order to promote the deposition of a well adhering, crystalline film it is
necessary to work at elevated temperatures or use other methods to introduce
additional energy into the reactive zone. The main advantages of CVD are that
complex surfaces can be coated and the film porosity can easily be controlled.
In the original (CVD) process a volatile inorganic Ti-compound (TiC1
) is
4
decomposed at the substrate in the presence of oxygen and! or water to Ti0
2
and HC1. The Ti-precursor is introduced into the reactor together with a carrier
gas, usually by bubbling it through the precursor fluid. Initially thermal energy
(heat) was used to promote the oxidation reaction and enhance film
crystallization. At temperatures between 300 to 700°C the anatase phase is
formed while at higher temperatures the rutile phase crystallizes. The exact
crystallization temperature depends on the nature of the process. Further
devëlopment of the CVD technique implemented other energy sources which
allow thin film crystallization at lower temperatures. Additionally, the carrier
gas flow and the concentration of the reactants influence the structure and
properties of the deposited film. Depending on the technique and the reaction
parameters used, quite high deposition speeds up to 10 nm/s can be reached. In
the following some variations of the standard thennal CVD technique will be
discussed briefly.
Since TiC1
4 is a quite unstable compound, investigators started using
organic Ti-compounds, i.e. metal-organic (MO) precursors, which has led to
the term MOCVD. The standard organic precursor is titanium tetra
isopropoxide (TTIP), also called Ti(IV)-isopropoxide, with the chemical
formula Ti(O-CH(CH
, which is normally implemented in form of a
4
2
)
3
solution in isopropanol. This material has a quite good stability towards
ambient air and a low toxicity so that it is easy to handle. Other organic Ti
compounds have also been used [72-75], inciuding precursors that do not need
addition of oxygen or water [76]. There is a continuous development of new
precursors with the aim to influence the deposition process and, as a result, the
film properties.
To enhance the dissociation of the precursors and or lower the reaction
temperature different secondary energy sources can be used. Examples are
(r.f.) plasma enhancement (PECVD) [77-87], thermal plasma (TPCVD) [88],
laser irradiation (LCVD) [89, 90] or ultraviolet light (UV) [91], with plasma
enhancement being the most implemented one.
Methods different than bubbling can also be used to introduce the Ti
precursor(s), e.g. liquid injection (LICVD) [91, 92], liquid delivery [93], liquid
mist spray [72] or liquid aerosol [94]. Related to the latter two techniques is
spray pyrolysis. The pressure in the reaction chamber can range from
atmospheric (APCVD) [71, 72, 95, 96] over low pressure [60, 84, 85] to high
vacuum [25].
214
K. Eufmger et al.
Atomic layer deposition
Further development of (M0)CVD has led to ALD (atomic layer
deposition) [97-101]. Here the reaction conditions and precursors are chosen in
a way as to yield a layer-by-layer growth of the film: First the Ti-precursor is
introduced and coated as a monolayer on the substrate, then an oxidizing agent
is introduced (e.g. 02), which oxidizes this layer. By repeating these steps a
(dense) thin film can be grown. Due to the low deposition rate, being on the
order of 1 0 nm!s, ALD is a very slow and therefore very expensive deposition
technique. The advantage is, though, that any surface shape can be coated in a
very homogeneous marmer, leading to so-called conformal coatings.
4.3.2. Physical vapor deposition techniques
Evaporation
Most evaporation techniques used for the deposition of Ti0
2 thin films are
derived from electron beam evaporation (Figure 9.). Here the source material,
usually Ti0
, a sub-oxide [102, 103] or Ti-metal [104] is heated with an
2
intense electron beam under an oxygen partial pressure below 0.1 Pa. The
addition of oxygen is needed to obtain stoichiometric films. In a standard setup
a low chamber pressure is needed to prevent scattering of the evaporated
particles and oxidation of the electron gun. 1f the electron gun is differentially
pumped higher pressures can be used in the deposition chamber. The
evaporation rates can be quite high, namely on the order of 10 nmls.
Crystalline (anatase or rutile) thin films can be obtained by either heating the
substrate during deposition or annealing after deposition. The microstructure of
the thin films can be controlled by the starting materials, the substrate
temperature [105, 106], the deposition (02) pressure [107, 108], the
evaporation rate and the angle of incidence [104]. Studies looking at a variety
of parameters are given in [102, 109]. An additional plasma [106] or ion
(beam) source [102, 105, 109, 110] can be implemented to control the thm film
microstructure and stoichiometry.
vacuum chamber
Figure 9. Scheme of a setup for electron beam evaporation.
2 thm films for photocatalytic applications
Ti0
215
Molecular beam epitaxy
Molecular beam epitaxy (MBE) is a very precise evaporation tecbnique
using an ultra low deposition speed. Here the Ti-precursor and the oxidizmg
agent are introduced together as atomic or molecular beams. The layers are
grown on a substrate which has a good lattice matching with the Ti0
2 phase to
be grown, so that the thin film follows the crystallographic orientation of the
substrate surface (epitaxial growth). In order to achieve this type of growth the
deposition rate bas to be rather low and the substrate needs to be chosen as to
yield a lattice match with the desired crystallographic orientation and phase of
. A very good review on the subject of MBE growth of oxide thin films,
2
Ti0
2 is given in [111]. Since MBE requires single
inciuding rutile and anatase Ti0
crystalline substrates, this is a technique which is not likely to be implemented
for large scale deposition of photocatalytic layers. Nevertheless, it is an
interesting technique for studying specific crystal planes of Ti0
.
2
Sputtering
In diode sputtering a discharge is generated by applying a d.c. voltage
between two electrodes (anode and cathode) in a low pressure gas, typically
argon (Figure 10, left). The ions generated in the discharge are accelerated
towards the cathode (target), bombarding its surface, which resuits in the
removal (sputtering) of cathode atoms [112, 113]. The sputter rate from a
diode discharge is low (partially due to the rather high pressures of 1-10 Pa
needed to sustain the discharge).
An improvement is the use of a magnetic field confinement of the
discharge above the cathode, i.e. magnetron sputtering [114, 115] (Figure 10,
right). This technique allows deposition at lower gas pressures (0.1 to 1.0 Pa)
and as a result, higher deposition rates. Further modifications are the use of r.f
and pulsed d.c. or a.c. power to avoid charging effects when sputtering from an
insulating target or in the presence of a reactive gas which forms a non-conductive
substrate: deposit thin film
electrons:
to anode
Figure 10. Scheme of d.c. diode (left) and magnetron (right) sputtering.
K. Eufmger et al.
216
compound with the sputtered target material. The deposition of Ti0
2 thin films
with (reactive) d.c. magnetron sputtering will be discussed in more detail in
section 5.
5. Deposition of Ti0
2 thin films by reactive d.c.
magnetron sputtering
In this chapter we will first give an overview of how the technique of
(reactive) d.c. magnetron sputtering can be used to deposit Ti0
2 thin films
(section 5.1). We will then discuss the parameters controllmg thin film growth
with this technique (section 5.2) and how they can be implemented for growing
2 thm films having different crystal and microstructures (section 5.3).
Ti0
5.1. Reactive d.c. magnetron sputtering
D.c. magnetron sputtering is a physical vapor deposition technique, which is
operated in a vacuum system at total gas pressures of 0.1-1 .OPa. 1f the sputtering
gas is inert, it will not undergo any chemical reaction with the target material so
that the composition of the thin film is controlled by the target composition.
When a second gas, which can react with the target material, is added, to the
sputtering gas the technique is called ‘reactive sputtering’ [116-118].
The addition of a reactive gas has profound implications on the sputtering
process and, therefore, on the deposition of the thin film. At low reactive gas
concentrations the same deposition conditions as for metallic sputtering (inert
gas only) are observed, so that metal rich thin films are deposited. For
intermediate reactive gas concentrations the process becomes unstable, so that
controlled thin film deposition in this range is not possible. For high reactive
gas concentrations the discharge conditions are quite different so that one
speaks about ‘reactive’ sputtering. In this regime a stoichiometric compound
Reactive gas: 02, N
2
sputter gas: Ari
target
powersupply
Figure 11. Scheme of a reactive d.c. magnetron sputtering setup.
2 thin films for photocatalytic applications
Ti0
217
film is deposited. For reactive sputtering the deposition rate is low, which is
unattractive for most applications. Therefore, methods have been developed to
stabilize the sputter process for intermediate reactive gas concentration which
allows depositing stoichiometric thin films at a higher rate [116-11 8].
In the case of Ti0
2 thin film deposition it is possible to use a sub
stoichiometric ceramic TiO
. target having a sufficient d.c. conductivity to
2
enable d.c. sputtering [22, 119-122]. In [123] the deposition rate of a Ti0
175
target in pure Ar is 50% of the metallic deposition rate and 3 times that of the
reactive rate. In [124] also a high deposition rate is reported for sputtering of a
TiO (1.5 <x< 1.7) target in pure Ar. In [119] the deposition rate of TiO
2 in
pure Ar is reported to be 6 times that of the rate in the reactive mode. It seems
that, when sputtered in pure Ar, the oxide target operates in the transition
region of the metallic target [125], and it is possible to deposit stoichiometric
films under these conditions [30, 126]. With increasing oxygen flow the same
sputtering conditions as for the metallic target in the reactive sputtering mode
are reached [125]. For TiO
.. targets no unstable sputtering regime is observed
2
[22, 122, 125], so that high rate deposition of stoichiometric films does not
require sophisticated process control [123, 124].
5.2. Structural development durmg film growth
A well known and much referred to model for the microstructural
development during thin film growth by metal evaporation is the one
developed Movchan and Demchishin [127], which was extended to sputtering
by Thornton [128]. Latter describes the changes in microstructure observed
when depositing thick (metallic) films with the substrate temperature T
(referred to the melting temperature Tm), and the inert gas pressure as
parameters. According to the observed thin film structure the growth was
divided into different zones. The Thomton model was further developed by
[129] relating the different zones to the particle energy instead of the surface
temperature. It is beyond the scope of our review to discuss this structure zone
model in detail, but we will give a summary here and point Out how it can be
used to define the deposition conditions for thin film structures that are of
interest for depositing photocatalytic Ti0
2 films.
In the structure zone model by [129] the thin film growth is divided into
three different growth zones (1-3), with an intermediate zone T between zone 1
and zone 2. In zone 1 the particle energy on the substrate surface is too low to
allow diffusion. For very low particle energies a so-called hit-and-stick growth
occurs, which resuits in amorphous or nanocrystalline, highly porous and
finely grained films. With increasing particle energy kinetic crystal growth can
occur in this zone 1, but the film density remains low. From zone T on the
arriving particles have enough energy to move on the substrate surface, so that
218
K. Eufmger et al.
preferential crystal growth can occur. As a result, the film density and the size
of the single crystal grains increase. In zone 3 the particle energy is high
enough to allow bulk diffusion, which should ultimately lead to the growth of a
single crystal film. The presence of defects and impurities prevents this due to
the formation of gram boundaries, though [129].
The important issues for depositing photocatalytic Ti0
2 are crystallinity,
stoichiometry (degree of oxidation), density and surface morphology. From the
description of the structure zones it becomes dear that zone 1, and to some
extent zone T, are the most interesting for photocatalytic Ti0
2 thin film
growth. The low density and small gram size will increase the surface area of
the thin film (section 3.6), enhancing its photocatalytic activity.
5.3. Controlling the structure of Ti0
2 thin films
As already discussed in section 3.1 the rutile phase is the
thermodynamically stable one, but since the energies of the anatase and
brookite phases are only slightly higher, they can exist as metastable phases up
to the transformation temperature of about 800°C [4]. For the gram sizes below
about 15 nm anatase is more stable than rutile or brookite, so that the crystal
phase formed can be controlled by the gram size.
The discussion of the thin film growth using the structure zone model
(section 5.2) becomes more complex in the case of Ti0
2 for two reasons: (i)
one has to take into account different crystal phases possible and (ii) the
sputtering process is reactive so that the oxygen partial pressure is an
additional parameter. In the following we give a short summary of the different
deposition conditions used to obtain rutile, anatase or amorphous Ti0
2 films
with reactive (d.c.) magnetron sputtering. No reference to the growth of
brookite films using (d.c.) magnetron sputtering was found.
Under most standard sputtering conditions and the substrate at room
temperature the particles forming the thin film remain quasi immobile on the film
surface so that the films follow the zone 1 type growth. This means that the Ti0
2
films are either XRD amorphous (section 3.2.1) or show only a very low degree
of crystallinity [130]. D.c. sputtering almost exclusively leads to the deposition
of the amorphous or the anatase structure, while pulsed d.c. and r.f. sputtering
can also yield rutile or mixed anatase/ rutile structures. For the latter techniques
the energy received by the growing film per incoming particle is higher at the
same input power, due to a lower deposition rate as well as a higher degree of
ionization in the plasma. As a result, a higher substrate self heating [114, 1311 is
observed. When crystalline films are deposited very often a preferential
orientation is observed, which may vary according to the deposition conditions.
The energy of the particles arriving at the substrate is mostly determined
by the sputter gas pressure and the target-to-substrate distance [114]. Both
2 thin films for photocatalytic applications
Ti0
219
control the extent of particle collision in their path between target and
substrate: Each collision leads to energy loss, and the number of collisions
increases with increasing pressure and target-to-substrate distance. Important is
also the oxygen partial pressure because (i) the deposition rate decreases with
increasing oxygen flow rate, remaining stable once the reactive mode is
established and (ii) the discharge voltage increases strongly upon transition
from the metallic to the reactive mode (when using metallic Ti targets). The
former decreases the energy received by the growing film per unit time while
the latter increases the energy per arriving particle. An additional factor is the
increase in substrate temperature either due to external heating or self heating
during the process. The total energy deposited in the thin film can additionally
be varied by the number of incoming particles per unit time. Latter is varied by
the sputter rate, which itself depends mostly on the discharge current [114].
Another method to increase the energy of the particles bombarding the thin
film is by using so-called’ unbalanced’ magnetron sputtering [129]. Here the
magnetic field lines direct the flux of charged particles present in the plasma to
the substrate, delivering additional energy to the growing film.
From the above discussion one can expect that rutile films are formed for
higher particle energies combined with a low deposition rate. Both result in
larger gram sizes, allowing the thermodynamically stable phase to form. This
is confirmed by experimental resuits for r.f., a.c. or pulsed d.c. sputtering given
in [132-136]. Very recently it was reported that an almost pure rutile film was
obtained using d.c. magnetron sputtering [137]. On the other hand, amorphous
or nanocrystalline films are expected to form at low particle energies and low
substrate temperatures, which is confirmed by experimental resuits
[32, 132, 135, 136, 138-140].
Anatase is reported to form at medium particle energies, especially when
smaller grains are formed. Elevated substrate temperatures will enhance
crystallization, but one needs to keep in mmd that temperatures above 800°C
almost exclusively result in the crystallization of rutile as well as strong gram
growth [4]. For r.f. sputtering the formation of anatase falis between the
conditions listed for rutile and amorphous Ti0
. For d.c. sputtering it was
2
found that the crystallization of anatase was promoted by increasing the
arriving particle energy [32, 138, 139-142]. The results indicate that there is no
difference using Ti and TiO
.. as target material.
2
5.4. Important issues for improving the photocatalytic activity
Up to now the highest photocatalytic activity has been reported for anatase
or anatase/ rutile mixed phases, so that the obvious aim is to deposit such
structures. The disadvantages of depositing crystalline thin films are the larger
gram size due to crystal growth and the elevated deposition temperatures
K. Eufmger et al.
220
needed. Our own studies have shown that amorphous thin films can have a
quite high photocatalytic activity when their porosity is larger than 20% [30].
Such films have a low gram diameter of 50 nm, which increases their active
surface area (section 3.6). Since they are deposited at temperatures below 50°C
it is possible to deposit them on heat sensitive substrates. Amorphous films can
be crystallized by an anneal treatment at temperatures between 300 and 400°C
[30, 143, 144]. These temperatures are too low for gram growth so that the fine
gram size of the as deposited coating is preserved.
6. Measuring the photocatalytic activity of Ti0
2 thin
films
In order to determine the activity of a photocatalyst it is necessary to have
an experimental procedure to either directly measure the conversion of light
energy into the lowering of the activation energy of a given breakdown
reaction or to compare the evolution of a chemical reaction between different
catalysts and! or organic substances. The former is rather difficult, if not
impossible to measure directly, while the latter gives only indirect evidence of
the catalyst activity. In this paragraph we point out the special considerations
for determining the photocatalytic activity of a thin film.
Figure 12. (left) shows the sketch of a generic setup for measuring the
photocatalytic activity of a thin film catalyst. The concentration of the organic
test substance (OR) and, if desired, of the decomposition product(s) (DP) are
followed by the analyzer for example a gas chromatograph (GC), a mass
spectrometer (MS) or a photo-spectrometer. Figure 12. (right) shows the
evolution of the OR and the DP for the case of first order reaction kinetics.
Normally the initial rate of the breakdown reaction (taken from the initial slope
of the concentration of the organic as a function of time) is used as a measure
for the catalytic activity of the thin film [145]. It can therefore be interesting
to change the order of the reaction to zero as this results in an initially linear
time of IIumnation
1
t
t
1
1
Figure 12. Sketch of a generic measurement setup to follow the photocatalytic
breakdown of an organic substance (OR) and its decomposition product (DP).
2 thm films for photocatalytic applications
Ti0
221
decay of the OR concentration [8]. It is a rather complex procedure to
determine the efficiency of a photocatalyst quantitatively. Standard definitions
and procedures were established by the IUPAC [145, 146] for the testing of the
photocatalytic activity of a powder suspended in water. These procedures are
designed to report the absolute photocatalytic activity of a given Ti0
,
2
referenced to the number of active photons (UV) arriving at the powder
surface.
Unfortunately the suggested reference organic compounds (phenol as
aromatic and formic acid as aliphatic organic) are not very suitable for testing
in a gaseous environment [145]. Additionally, the true surface area of a thin
film is usually not measurable (section 3.6.1), even though it can be estimated
using the procedure outlined in section 3.6.2. As a result, most experimental
settings for testing thin film photocatalysts are not compatible with the strict
requirements stated by the ILTPAC, meaning that a more qualitative approach
must be applied. One option is to use a reference thin film catalyst against
which other thin films, organic substances or changes in the experimental
settings can be tested. The major problem is, though, that no universally
2 thin film catalyst material exists. The only reports of a reference
available Ti0
thin film were found in [47, 147], where a spin coated layer of the well-known
powder catalyst Degussa P25 was used. A second suggested material was
Pilkington ActiveTM [47], a CVD deposited thin film on Si0
2 coated glass, but
this material has a very low photocatalytic activity due to is low thickness [31].
There are several important parameters when testing and comparing thin
film photocatalysts. The first is the number of (active) photons (i.e. the IJV
intensity) absorbed in the thin film. This number depends on the film thickness
(section 3.7), the spectral range and power of the (UV) lamp used and how
much light is lost between lamp and thin film surface. The second parameter is
the initial concentration of the organic test substance in relation to the active
thin film surface area, which determines the coverage of the catalyst surface
and therefore the reaction kinetics [4, 8]. The third parameter is the nature of
the organic test substance, which determines the active pore size and therefore
the active surface area of the thin film. Additional parameters are the degree of
surface hydroxylation (section 3.4) and the presence of oxygen. Latter is
normally taken care of by working in air or by supplying a surplus of oxygen
to the reaction chamber (e.g. by bubbling).
7. Conciusions and outlook
In this review we discussed several aspects of TiO
2 thin film deposition fQ.r
photocatalytic applications. We started with the factors influencing the
2 thin films, the two most important being the
photocatalytic activity of Ti0
microstructure and the crystallinity. Even though the active surface area is one
K. Eufmger et al.
222
of the most important factors for a catalyst, the effect of the microstructure on
the photocatalytic activity is neglected in most studies on thin films. In this
review we demonstrated how the microstructure (porosity) of the thin films
can enlarge the surface area. Next to keeping the gram size small it is
important to estimate the pore size of the thin film and to determine whether
these pores are large enough to enable inward diffusion of the organic
substance studied and outward diffusion of reaction products. Additionally the
thickness of the thin film is important, especially for high porosity, because the
depth of the pores and, as a result, the active film surface area increases.
The thin film crystallinity is the other important factor. Crystalline defects
can act as trap sites for the photogenerated charge carriers, reducing the
number of charge carriers available for the reaction to be catalyzed. For high
crystallinity it is interesting to minimize the number of gram boundaries, i.e. to
increase the gram size. This is contrary to the desire to increase the surface
area, so that usually a gram size range of about 10-20 nm is considered an
optimum compromise between these two issues. From the three most common
2 crystalline phases anatase is reported to be the most photoactive one. This
Ti0
is one of the reasons why most studies work with crystalline anatase thin films.
Anatase is also the crystal phase which forms most easily for small gram sizes
and at the processing conditions commonly used.
2 thin films can be deposited with a variety of techniques based on wet
Ti0
chemical and vapor deposition techniques. The most popular ones are sol-gel,
chemical vapor deposition (CVD), and magnetron sputtering, a physical vapor
deposition (PVD) technique. Independent of the deposition technique used it is
desirable to control the deposition parameters as to yield thin films that fulfill
the two basic requirements of high surface area and good crystallinity. For
many techniques crystalline thin films can only be obtained when heated
during or after deposition. This limits the substrate material choice, with most
polymers not being applicable. Additionally, uptake of sodium or other
detrimental contaminants from the substrate must be avoided. An altemative
can be to deposit a thm film of low crystallinity at low temperatures and
optimize its microstructure as to yield a high surface area.
Unfortunately there is no standard procedure or reference sample for
measuring the photocatalytic activity of Ti0
2 thin films. This makes it difficult
to compare results between laboratories or different measurement setups. We
think that this issue is an important one for the future.
8. References
1.
2.
3.
Fujishima A., Honda K. 1972, Nature 238, 37.
Schiavello M. 1997, Heterogeneous photocatalysis, Wiley Series in Photoscience
and Photoengineering, Volume 3, John Wiley & Sons, Chichester.
Anpo M., Takeuchi M. 2003, J. Catal. 216, 505.
2 thm films for photocatalytic applications
Ti0
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
24.
25.
26.
27.
28.
29.
30.
31.
32.
33.
34.
223
Carp 0., Huisman C.L., Reller A. 2004, Prog. Solid State Chem. 32, 33.
Fernândez-Garcia M.,. Martinez-Arias A, Hanson J.C., Rodriguez J.A. 2004,
Chem. Rev. 104, 4063.
Fujishima A., Rao T.N., Tryk D.A. 2000, J. Photochem. Photobio. C: Photochem.
Rev. 1, 1.
Linsebigler A.L., Lu G., Yates Jr J.T. 1995, Chem. Rev. 95, 735.
Muis A., Le Hunt S., J. 1997, Photochem. Photobio. A: Chem. 108, 1.
Karches M., Morstein M., von Rohr P.R., Pozzo R.L., Giombi J.L., Baltanâs M.A.
2002, Catal. Today 72, 267.
Boulesteix C., Kang Z., Lottiaux M. 1986, Phys. Stat. Sol. (a) 94, 499.
Takeda S., Suzuki S., Odaka H., Hosono H. 2001, Thin Solid Films 392, 338.
Nam H.-J., Amemiya T., Murabayashi M., Itoh K. 2004, J. Phys. Chem. B
108, 8254.
Fallet M., Permpoon S., Deschanvres J.L., Langlet M. 2006, J. Mater. Sci.
41, 2915.
Smith D.L. 1995, Thin Film Deposition, Principles and Practice, Chap. 5:
Deposition, McGraw-HilI, mc., New York, 119.
Eufinger K. 2007, PhD Thesis, Ghent University, Ghent, Belgium.
Damm C., Völtzke D., Abicht H.-P., Israel G. 2005, J. Photochem. Photobio. A:
Chem. 174, 171.
Yin S., Hasegawa H., Maeda D., Ishitsuka M., Sato T. 2004, J. Photochem.
Photobio. A: Chem. 163, 1.
Miyagi T., Kamei M., Mitsuhashi T., Ishagashi T., Yamazaki A. 2004, Chem.
Phys. Lett., 390, 399.
Ohno T., Sarukawa K., Tokieda K., Matsumura M. 2001, J. Catal. 203, 82.
Ohno T., Tokieda K., Higashida S., Matsumara M. 2003, Appi. Cat. A: General
244, 383.
Zailen R., Moret M.P. 2006, Solid State Commun. 137, 154.
Kominami H., Ishii Y., Kohno M., Konishi S., Kera Y., Ohtani B. 2003, Catal.
Lett. 91, 41.
J.I. Pankovel97l, Optical processes in semiconductors, Dover Publications, mc.,
New York.
Jung K.Y., Park S.B. 1999, J. Photochem. Photobio. A: Chem. 127, 117.
Jung K.Y., Park S.B., Ibm S.-K. 2002, Appi. Cat. A: General 224, 229.
Jung K.Y., Park S.B. 2004, Mater. Lett. 58, 2897.
Baroch P., Musil J., Vicek J., Nam K.H., Han J.G. 2005, Surf. Coat. Technol.
193, 107.
Hove M.A. 2006, Catal. Today 113, 133.
Zeman P., Takabayashi S. 2003, Thin Solid Films 433, 57.
Eufinger K., Poelman D., Poelman H., De Gryse R., Marin G.B. 2007, Vac.
Technol. Coat. 8, 44.
Eufinger K., Poelman D., Poelman H., De Gryse R., Marin G. B. 2007, J. Phys. D:
Appi. Phys. 40, 5232.
Liu B., Zhao X., Zhao Q., Li C., He X. 2005, Mater. Chem. and Phys. 90, 207.
Zhang W., Li Y., Zhu S., Wang F. 2004, Chem. Phys. Lett. 373, 333.
Asahi R., Morikawa T., Ohwaki T., Aoki K., Taga Y. 2001, Science 293, 269.
224
K. Eufmger et al.
35. Ihara T., Miyoshi M., Iriyama Y., Matsumoto 0., Sugihara S. 2003, AppI. Cat. B:
Environ. 42, 403.
36. Prabakar K., Takahashi T., Nezuka T., Nakashima T., Kubota Y., Fujishima A.
2006, J. of Vac. Sci. and Technol. A 24, 1156.
37. Miyagi T., Kamei M., Mitsuhashi T., Yamazaki A. 2006, AppI. Phys. Lett. 88,
Artn 132101.
38. Wang Y., Doren D.J. 2005, Solid State Commun. 136, 186.
39. Mathieu H., Pascual J., Camassel J. 1978, Phys. Rev. B 18, 6920.
40. Miao L., Tanemura S., Watanabe H., Toh S., Kaneko K. 2005, Appi. Surf Sci.
244, 412.
41. Wunderlich W., Oeckermann T., Miao L., Hue N.T., Tanemura S., Tanemura M.
2004, J. Ceram. Process. Res. 5, 343.
42. Wunderlich W., Miao L., Tanemura M., Tanemura S., Jin P., Kaneko K., Terai A.,
Nabatova-Gabin N., Belkada R. 2004, Int. J. Nanosci. 3, 439.
43. Anderson 0., Ottermann C.R., Kuschnereit R., Hess P., Bange K. 1997, Fresenius
J. Anal. Chem. 358, 315.
44. Fernândez A., Lassaletta G., Jiménez V.M., Justo A., Gonzâlez-Elipe A.R.,
Herrmann J.-M., Tahiri, H. Ait-Ichou Y. 1995, Appi. Cat. B: Environ. 7, 49.
45. Tada H., Tanaka M. 1997, Langmuir 13, 360.
46. Tomaszewski H., Eufinger K., Poelman II., Poelman D., De Gryse R., Smet P.F.,
Marin G.B. 2007, Int. J. Photoenergy 8, ID 95213.
47. Muis A., Lepre A., Elliot N., Bhopal S., Parkin I.P., O’NeilI S.A. 2003, J.
Photochem. Photobio. A: Chem. 160, 213.
48. Guan K. 2005, Surf Coat. and Technol. 191, 155.
49. Sakthivel S., Hidalgo M.S., Bahnemann D.W., Geissen S.-U., Murugesan V.,
Vogelpohi A. 2006, Appi. Cat. B: Environ. 63, 31.
50. Cornu C.J.G. 2002, PhD Thesis, California Institute of Technology, Pasadena,
California.
51. Tjong S.C., Chen H. 2004, Mater. Sci. Eng. R 45, 1.
52. Li W., Ismat Shah S., Huang C.-P., Jung 0., Ni C. 2002, Mater. Sci. Eng. B
96, 247.
53. Dumitriu D., Bally A.R., Ballif C., Hones P., Schmid P.E., Sanjinés R., Lévy F.,
Pârvulescu V.I. 2000, Appi. Cat. B: Environ. 25, 83.
54. Madhusudan Reddy K., Manorama Sunkara V., Ramachandra Reddy A. 2002,
Mater. Chem. Phys. 78, 239.
55. Monticone S., Tufeu R., Kanaev A.V., Scolan E., Sanchez C. 2000, Appi. Surf.
Sci. 162-163, 565.
56. Baklanov M.R., Moglinov K.P., Polovinkin V.G., Dultsev F.N. 2000, J. Vac. Sci.
Technol. B 18, 1385.
57. Poelman D., Smet P.F. 2003, J. Phys. D: Appi. Phys. 36, 1850.
58. http://en.wikipedia.org/wiki/Mesopore, /Macropore, /Micropore.
59. Eufïnger K., Poelman D., Poelman H., De Gryse R., Marin G. B. 2007, Appi. Surf.—
Sci. 254, 148.
60. Jung S.-C., Kim S.-J., Imaishi N., Cho Y.-I. 2005, Appi. Cat. B: Environ. 55, 253.6
61. Kern P., Schwalier P., Michier J. 2006, Thin Solid Films 494, 279.
2 thm films for photocatalytic applications
Ti0
225
62. An H.J, Jang S.R., Vittal R., Lee J., Kim K.J. 2005, Electrochim. Acta, 50, 2713.
63. Manivannan A., Spataru N., Arihira K., Fushishima A. 2005, Electrochem. Solid
State Lett. 8, 138.
64. Georgieva J., Armyanov S., Valova E., Poulios 1., Sotiropoulos S. 2006,
Elechtrochim. Acta 51, 2076.
65. Yan H.W., Yang Y.L., Fu Z.P., Yang B.F., Xia L.S., Xu Y.D., Fu S.Q., Li F.Q.
2006, Chem. Lett. 35, 864.
66. Karuppuchamy S., Jeong J.M., Amalnerkar D.P., Mmoura H. 2006, Vacuum 80, 494.
67. IshikawaY., Matsumoto Y. 2002, Solid State Ionics 151, 213.
68. Wessels K., Feldhoff A., Wark A., Rathhousky J., Oekermann T. 2006,
Elechtrochem. Solid-State Lett. 9, C93.
69. Sankapal B.R., Sartale S.D., Lux-Steiner M.C., Ennaoui A. 2006, C.R. Acad. Sci.,
Ser. lic: Chim. 9, 702.
70. Zotti G., Schiavon G., Zeechin S. 1999, J. Electrochem. Soc. 146, 637.
71. Backman U., Auvinen A., Jokiniemi J.K. 2005, Surf. Coat. Technol. 192, 81.
72. Kim B.H., Lee J.Y., Choa Y.H., Higuchi M., Mitzutani N. 2004, Mater. Sci. Eng.
B 107, 289.
73. Woo K., Lee W.I., Lee J.S., Kang S.O. 2003, Inorg. Chem. 42, 2378.
74. Sato N., Nakajima K., Usami N., Takahashi H., Muramatsu A., Matsubara E. 2002,
Mater. Transact. 43, 1533.
75. Besserguenev V.G., Pereira R.J.F., Mateu M.C.s, Khmelinskii I.V., Nicula R.C.,
Burkel E. 2003, Int. J. Photoenergy 5, 99.
76. Jung C.K., Lee S.B., Boo J.H., Ku S.J., Yu K.S., Lee J.W. 2003, Surf. Coat.
Technol. 174, 296.
77. Amassian A., Svec M., Desjardins P., Martinu L. 2006, J. Vac. Sci. Technol. A
24, 2061.
78. Horakova M., Kolouch A., Spatenka P. 2006, Chech. J. Phys. 56, Bi 185.
79. Kambe M., Fukawa M., Taneda N., Sato K. 2006, Sol. Engergy Mater. Sol. Ceils
90, 3014.
80. Borras A., Barranco A., Gonzalez-Elipe A.R. 2006, J. Mater. Sci. 41, 5220.
81. Sonnenfeld A., Hauert R., von Rohr P.R. 2006, Plasma Chem. Plasma Process.
26, 319.
82. Maeda M., Watanabe T. 2005, Thin Solid Films, 489, 320.
83. Hatanaka Y., Naito N., Itou S., Kando M. 2005, Appi. Surf: Sci. 244, 554.
84. AhnK.H., ParkY.B., ParkD.W. 2003, Surf Coat. Technol. 171, 198.
85. Huang S.S., Chen J.S. 2002, J. Mater. Sci. Maters Eletron. 13, 77.
86. Nakamura M., Kato S., Aoki T., Sirghi L., Hatanaka Y. 2001, Thin Solid Films
401, 138.
87. Martinu L., Poltras D. 2000, J. Vac. Sci. Technol. A 18, 2619.
88. Ando Y., Tobe S., Tahara H. 2006, Vacuum 80, 1278.
89. Goto T., Kimura T. 2006, Science of Engineering Ceramics III, Series on Key
Engineering Materials, 317-3 18, 495.
90. Halary-Wagner E., Wagner F.R., Brioude A., Mugnier J., Hoffmann P. 2005,
Chem. Vap. Deposition 11, 29.
91. Zhang J.Y., Boyd I.W., O’Sullivan B.J., Hurley P.K., Kelly P.V., Senateur J.P.
2002, J. Non-Cryst. Sol. 303, 134.
-
K. Eufinger et al.
226
92. Simcock M.N. 2006, Surf. Interface Mal. 38, 1122.
93. No S.Y., Oh J.H., Jeon C.B., Schindler M., Hwang C.S., Kim H.J. 2005, J.
Electrochem. Soc. 152, C435.
94. Conde-Gallardo A., Guerrero M., Fragoso R., Castillo N. 2006, J. Mater. Res.
21, 3205.
95. Nolan M.G., Pembie M.E., Sheel D.W., Yates H.M. 2006, Thin Solid Films
515, 1956.
96. Richards B.S., Huong N.T.P., Crosky A. 2005, 3. Electrochem. Soc. 152, F71.
97. Lim G.T., Kim D.H. 2006, Thin Solid Films 498, 254.
98. Kim S.K., Kim K.M., Kwon O.S., Lee S.W., Jeon C.B., Park W.Y., Hwang C.S.
2005, J. Jeong, Electrochem. Solid-State Lett. 8, 59.
99. Pore V., Rathu A., Leskela M., Ritala M., Sajavaara D., Keinonen J. 2004, Chem.
Vap. Deposition 10, 143.
100. Mitchel D.R.G., Attard D.J., Triani G. (2003), Thin Solid Films 441, 85.
101. Schuisky M., Aarik J.,Kukli K., Aidla A., Harsta A. 2001, Langmuir 17, 5508.
102. Rao K.N. 2002, Opt. Eng. 41, 2357.
103. Selhofer H., Ritter E., Linsbod R. 2002, Appi. Opt. 41, 756.
104. Li C.R., Zheng Z.H., Zhang F.M., Yang S.Q., Wang H.M., Chen L.Z., Zhang F.,
Wang X.H., Liu X.H. 2000, Nuci. Instr. Method. Phys. Res. B 169, 21.
105. Chen H.C., Lee C.C., Jaing C.C., Shiao M.H., Lu C.J., Shieu F.S. 2006, Appi. Opt.
45, 1979.
106. Modes T., Scheffel B., Metzner C., Zywitzki 0., Reinhold E. 2005, Surf. Coat.
Technol. 200, 306.
107. Jiang J.C., Li B., Zhang Z.Q., Hu P.G., Zhang F.S., Cheng H.S., Yang F.3. 2002,
Phys. Sta. Sol. (a) 193, 69.
108. Yang T.S., Shiu C.B., Wong M.S. 2004, Surf. Sci. 548, 75.
109. Tien C.L., Lin S.W. 2006, Opties. Commun. 266, 574.
110. Guo A.Y., Xue Y.Y., Zhu X.M., Zhang G.Y., Guo P.T., Hu X.F. 2006, J. Wuhan.
Univ. Technol. Mater. Sci. Ed. 21, 101.
111. Chambers S.A. 2000, Surf. Sci. Rep. 39, 105.
112. Chapman B.N. 1980, Glow Discharge Processes, John Wiley & Sons, New York.
113. Vossen J.L., Cuomo J.J. 1978, Thin film processes Pt. 2, chapter 11-1, Vossen J.L.,
Kern W. (ed.), Academic Press.
114. Westwood W. 2003, Sputter deposition, AVS Education Committee Book Series,
Volume 2.
115. Waits R.K. 1978, Thin film processes Pt. 2, chapter 11-4, Vossen J.L., Kern W.
(ed.), Academie Press.
116. Berg S., Nyberg T. 2005, Thin Solid Films 476, 215.
117. Safi 1. 2000, Surf. Coat. Technol. 127, 203.
118. Sproul W.D., Christy D.J., Carter D.C. 2005, Thin Solid Films 491, 1.
119. Ohsaki H., Tachibana Y., Hayashi A., Mitsui A., Hayashi Y. 1999, Thin Solid
Films, 351, 57.
120. Ohsaki H., Tachibana Y., Mitsui A., Kamiyama T., Hayashi Y. 2001, Thin Solid
Films 392, 169.
121.Tachibana Y., Ohasaki H., Hayashi A., Mitsui A., Hayashi Y. 2000, Vacuum
59, 836.
-
2 thin films for photocatalytic applications
Ti0
227
122. Tomaszewski H., Poelman H., Depla D., Poelman D., De Gryse R., Fiermans L.,
Reyniers M.F., Heynderickx G., Marin GB. 2003, Vacuum 68, 31.
123. Poelman H., Tomaszewski, H. Poelman D., Depla D., De Gryse R. 2004, Surf.
Interface Anal. 36, 1167.
th
124. Persoone P., Diericks S., Luys S., De Bosscher W. 2005, Proc. 48
Ann. Tech.
Conf. Soc. of Vacuum Coat. (SVC), Denver, Co, 180.
125. Depla D., Tomaszewski H., Buyle G., De Gryse R. 2006, Surf. Coat. Technol.
201, 848.
126. Eufinger K., Janssen E.N., Poelman H., Poelman D., De Gryse R., Marin G.B.
2006, Thin Solid Films 515, 425.
127. Monchavan B.A., Demchishin A.V. 1969, Phys. Met. Metallogr. 28, 83.
128. Thornton J. A. 1977, Ann. Rev. Mater. Sci. 7, 239.
129. Mahieu S., Ghekiere P., Depla D., De Gryse R. 2006, Thin Solid Films 515, 1229.
130. Baroch P., Musil J., Vlëek J., Nam K.H., Han J.G. 2005, Surf. Coat. Technol.
193, 107.
131. Snyders R., Gouttebaron R., Dauchot J.P., Hecq M. 2005, Surf. Coat. Technol.
200, 448.
th
132. Asanuma T., Matsutani T., Liu C., Mihara T., Kiuchi M. 2003, Proc. 7
Int. Symp.
Sputt. Plasma Proc. (ISSP 2003), Kanazawa Inst. Technol., Kanazawa, Japan, 447.
133. Miao L., Jin P., Kaneko K., Tanemura S. 2002, Proc. Advanced Nanomater.
Nanodevices (IUMRS-ICEM), Xi’an, China, 943.
134. Okimura K. 2001, Surf. Coat. Technol. 135, 286.
135. Yamagishi M., Kuriki S., Shigesato Y. 2003, Thin Solid Films 442, 227.
136. Zeman P., Takabayashi S. 2002, Surf. Coat. Technol. 153, 93.
137. Vale A., Chaure N., Simonds M., Ray A.K., Brickiebank N. 2006, J. Mater. Sci.
Mater. Electron. 17, 851.
138. Bames M.C., Kumar S., Green L., Hwang N.-M., Gerson A.R. 2005, Surf. Coat.
Technol. 190, 321.
139. Kim S.-H., Choi Y.-L., Song Y.-S., Lee D. Y., Lee S.-J. 2002, Mater. Lett.,
57, 343.
140. Takahashi T., Prabakar K., Nezuka T., Yamazaki T., Nakashima T., Kubota Y.,
Fujishima A. 2006, J. Vac. Sci. Technol. A 24, 1161.
141.Zhang W., Li Y., Zhu S., Wang F.2004, Surf. Coat.s Technol. 182, 192.
142. Takahashi T., Nakabashi H., Mizuno W. 2002, J. Vac. Sci. Technol. A 20, 1916.
143. Martin N., Rousselot C., Rondot D., Palmino F., Mercier R. 1997, Thin Solid
Films 300, 113.
144. Zheng S.K., Wang T.M., Xiang G., Wang C. 2001, Vacuum 62, 361.
145. Serpone N., Salinaro A. 1999, Pure Appi. Chem. 71, 303.
146. Salinaro A., Emeline A.V., Zhao J., Hidaka H., Rybabchuk V.K., Serpone N. 1999,
Pure Appi. Chem. 71, 321.
147. Muis A., Elliot N., Parkin I.P, O’Neill S., Ciark R. J. 2002, J. Photochem.
Photobio. A: Chem. 151, 171.
-